A Lightweight Visual Programming tool for Machine Learning and Data Manipulation

In recent decades, visual programming tools have focused on scientific workflow tasks. However, the current focus has shifted to machine learning tasks, which are most often amenable to unification. The existing solutions in this area demonstrate many alternatives with the possibility of using model...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2020 International Conference on Computational Science and Computational Intelligence (CSCI) s. 981 - 985
Hlavní autoři: Khodnenko, Ivan, Ivanov, Sergey V., Prokofiev, Kirill, Lantseva, Anastasiia
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.12.2020
Témata:
Computational modeling
Data models
dataflow programming
Machine learning
Programming
Scientific computing
visual programming
Visualization
workflows
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In recent decades, visual programming tools have focused on scientific workflow tasks. However, the current focus has shifted to machine learning tasks, which are most often amenable to unification. The existing solutions in this area demonstrate many alternatives with the possibility of using models from various machine learning packages. The proposed solution is based on the concept of minimal graphical notation, where all typical operations on data and machine learning models are possible. Another distinguishing feature of the proposed solution is the explicit separation of data and models, making the proposed notation clearer for perception. In the proposed engine, computational graph traversal does not have a rigorous sequence. The call of the sequence of calculations depends on the specific model that the user currently needs.
DOI:10.1109/CSCI51800.2020.00182